13:59:58 <nikhil> #startmeeting glance 13:59:59 <openstack> Meeting started Thu Apr 7 13:59:58 2016 UTC and is due to finish in 60 minutes. The chair is nikhil. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:01 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:03 <openstack> The meeting name has been set to 'glance' 14:00:08 <nikhil> #topic roll call 14:00:13 <nikhil> o/ 14:00:15 <mfedosin> o/ 14:00:17 <bunting> o/ 14:00:33 <rosmaita> o/ 14:00:43 <hemanthm> o/ 14:01:12 <Steap> o/ 14:01:22 <avarner> o/ 14:01:36 <nikhil> let's get started 14:01:38 <nikhil> #topic agenda 14:01:44 <kairat> o/ 14:01:44 <nikhil> we've a short agenda today 14:02:05 <nikhil> I want to spend as much time as possible on discussing summit sessions 14:02:15 <nikhil> https://etherpad.openstack.org/p/glance-team-meeting-agenda 14:02:35 <nikhil> #topic Updates 14:02:41 <tjcocozz> o/ 14:02:45 <nikhil> #info Glare updates ( mfedosin ) 14:02:53 <mfedosin> hey! 14:03:06 <nikhil> hi Mike 14:03:11 <nikhil> how are things? 14:03:20 <nikhil> (in the artifacts world) 14:03:33 <mfedosin> we spent last three days discussing possible improvements of Glare architecture 14:04:01 <mfedosin> 1. we simplified our engine 14:05:08 <mfedosin> fyi engine is a thing that does policies check, notifications and access control 14:05:19 <dshakhray> o/ 14:05:45 <mfedosin> code was reduced twice there 14:06:02 <mfedosin> 2. we rethought how we delete artifacts 14:06:29 <mfedosin> if you read the spec deletion was rather complicated there 14:06:36 <mfedosin> we decided to keep it simple 14:07:02 <mfedosin> first of all we got rid of 'killed' and 'deleted' statuses for blobs 14:07:18 <mfedosin> and deletion looks like: 14:07:54 <mfedosin> 1. Glare set status 'deleted' to artifact and 'pending_delete' to all his blobs 14:08:09 <mfedosin> it's a transaction operation 14:08:39 <mfedosin> 2. Then Glare starts to remove files from store and remove blob instances from db in cycle 14:09:41 <mfedosin> If something bad happens like lost connection with data storage then glare raises exception and leaves other blobs in penging_delete status 14:09:58 <mfedosin> and operator can clean the storage with scrubber 14:10:42 <mfedosin> If delayed delete is activated then only 1 step is performed and Glare misses the second 14:11:03 <mfedosin> that's all updates we have from Glare 14:11:09 <mfedosin> thanks :) 14:11:20 <nikhil> Thanks mfedosin 14:11:34 <nikhil> Are you wanting to get feedback on any of these items? 14:11:55 <nikhil> guess, that's a no 14:12:00 <mfedosin> I'm not sure if it's the best place to get the feedback 14:12:11 <hemanthm> mfedosin: just curious, how do you ensure #1 is a transaction? 14:12:20 <hemanthm> (feel free to say we can take it offline) 14:12:21 <mfedosin> if you have something to say feel free to join glare meeting on Monday 14:12:32 <nikhil> mfedosin: sure, I was saying if we need to link something here so that people can go and give that feedback 14:12:49 <mfedosin> hemanthm: it's a regular update 14:12:56 <mfedosin> one call to db 14:12:59 <hemanthm> two updates? 14:13:01 <hemanthm> ok 14:13:23 <nikhil> ok, mfedosin can link the spec or feedback form later 14:13:27 <nikhil> moving on 14:13:33 <nikhil> #info Nova v1, v2 updates 14:13:34 <mfedosin> set status 'deleted' to artifacts and 'pending_delete' to its blobs 14:13:59 <mfedosin> it's me again :) 14:14:17 <mfedosin> so, I proposed a spec for this update 14:14:36 <mfedosin> #link https://review.openstack.org/#/c/301741/ 14:14:55 <mfedosin> there is huge feedback 14:15:28 <nikhil> yes and some more feedback is on this email thread 14:15:32 <nikhil> #link https://www.mail-archive.com/openstack-dev@lists.openstack.org/msg79862.html 14:15:32 <mfedosin> also I send an email to ML where I describe things that are done 14:15:43 <mfedosin> and should be done 14:16:02 <nikhil> #link https://www.mail-archive.com/openstack-dev@lists.openstack.org/thrd2.html#79862 14:16:10 <mfedosin> hope to find some time today to answer there 14:16:17 <mfedosin> thanks nikhil 14:16:22 <nikhil> np 14:16:33 <nikhil> I think we've a good momentum on the work 14:16:58 <nikhil> Also, the thread has some context that I got from irc conversation on -nova 14:17:17 <nikhil> the proposed changes look like something that can be done in specless BP 14:17:55 <mfedosin> em... what? 14:17:59 <nikhil> We should ad-sync on Monday (decide on a time offline) on this topic (whomever interested) 14:18:24 <mfedosin> I think spec is needed there 14:18:29 <nikhil> in a gist, the discovery related changes 14:19:02 <nikhil> I don't want to say what nova spec is needed or not needed, if possible I'd refrain from proposing another nova spec 14:19:26 <nikhil> let's try to get things done in a specless BP for things that are not too broad in changes proposed 14:19:56 <nikhil> I can update more stuff on -glance 14:20:02 <mfedosin> hmm... we need to discuss it later 14:20:04 <nikhil> moving on 14:20:08 <nikhil> #topic Releases 14:20:10 <mfedosin> nikhil: yes 14:20:51 <nikhil> THere were some release notes changes that release team was trying to get in 14:21:22 <nikhil> if you see a holler on #openstack-glance please courtesy them with a review. Thanks in advance. 14:21:42 <nikhil> that's it on this on 14:21:45 <nikhil> moving on 14:21:51 <nikhil> #topic Announcements 14:21:59 <nikhil> Two things 14:22:28 <nikhil> #info Glance virtual pre summit sync is on Tuesday April 12 1400-1800 UTC 14:22:44 <nikhil> we will be using the same etherpad for summit planning viz. 14:22:56 <nikhil> #link https://etherpad.openstack.org/p/newton-glance-summit-planning 14:23:18 <nikhil> if someone needs to discuss a CP topic, we may be able to give short feedback (~20 mins) 14:23:37 <nikhil> #link https://etherpad.openstack.org/p/newton-cross-project-sessions 14:23:47 <nikhil> those are in the process of being selected 14:24:03 <nikhil> second 14:24:33 <nikhil> #info Glance Newton-mid-cycle will be in Boston (we're finalizing on the dates with the sponsors) 14:24:50 <mfedosin> when? 14:24:52 <nikhil> Hopefully that will give you some time to start planning travel 14:25:07 <nikhil> It will be sometime in mid-June 14:25:28 <nikhil> final dates will be announced once the sponsors give thumbs up 14:25:51 <nikhil> any questions or other announcements? 14:26:01 <mfedosin> also there was a doodle afair 14:26:32 <nikhil> yes, that was to get feedback from the team. sponsors are on their independent timeline :) 14:26:56 <nikhil> if you're interested in knowing 14:27:01 <nikhil> here's the tentative dates 14:27:03 <nikhil> #link http://doodle.com/poll/2c6dqs9ab253a59m 14:27:35 <mfedosin> it also show who's going to attend it 14:27:47 <nikhil> s/is/may be/ 14:28:36 <nikhil> we will have a more concrete agenda for mid cycle closer to the event 14:28:40 <nikhil> this was done to help plan travel 14:29:08 <nikhil> that document will be more accurate of the anticipated participation, schedule and related events during the meetup 14:29:21 <nikhil> moving on 14:29:29 <nikhil> (you can ask me ques offline) 14:29:33 <nikhil> #topic Glance summit sessions continued 14:29:42 <nikhil> #link https://etherpad.openstack.org/p/newton-glance-summit-planning 14:30:02 <nikhil> SO, the continuation is from this part 14:30:04 <nikhil> #link http://eavesdrop.openstack.org/meetings/glance/2016/glance.2016-03-24-13.59.log.html#l-112 14:31:03 <nikhil> we've discussed proposals 3, 4, 8, 9 & 16 as of now 14:31:40 <nikhil> oops 14:31:50 <nikhil> 15 too 14:32:23 <nikhil> rosmaita: I am putting you on the spot here to discuss proposal #14 if you 14:32:26 <nikhil> you 14:32:28 <nikhil> eh 14:32:32 <rosmaita> sure 14:32:33 <nikhil> keyboard fail 14:32:54 <nikhil> #info discussion on Proposal #14 ( rosmaita ) 14:33:02 <rosmaita> two related issues about image properties 14:33:25 <rosmaita> one is whether glance shoudl enforce metadef-defined property values on images 14:33:37 <rosmaita> the other is whether we can come up with a lifecycle for public images 14:33:40 <rosmaita> in this sense 14:34:00 <rosmaita> deployers create image 1, end-users boot from it 14:34:16 <rosmaita> somebody finds an issue with image 1, so the deployer creates image 2 14:34:35 <rosmaita> all new people who want to use that operating system should use image 2 14:34:50 <rosmaita> but some people may still want to use image 1 and patch by themselves 14:35:06 <rosmaita> so the problem is: how to "hide" image 1 but have it still be public 14:35:33 <rosmaita> and obviously this will be a series of >2 images for the lifecycle of a particular operating system image 14:35:34 <hemanthm> image deprecation :) 14:36:22 <rosmaita> there's been interest in this kind of thing for a while, various clouds have their own hacks to make it work, but it seems like a real use case that we should actually provide a solution for 14:36:42 <rosmaita> someone was in the glance channel earlier this week asking about this 14:36:54 <hemanthm> rosmaita: so these older images are expected to stay around forever? 14:37:23 <rosmaita> hemanthm: yes 14:37:43 <nikhil> rosmaita: are we trying to present a set of formal (protected) properties for this? 14:38:06 <rosmaita> i think so 14:38:09 <nikhil> I guess your problem statement is quite clear, I'm curious about the title 14:38:11 <rosmaita> actually, i am not sure 14:38:16 <rosmaita> hence the discussion 14:38:28 <nikhil> ok, so this will be something that will revolve around feedback? 14:38:45 <rosmaita> unless we can come up with somethign that we think will work pre-summit 14:38:52 <rosmaita> (that's what i'd prefer) 14:39:04 <nikhil> I am trying to collect this info to tag potential particular sessions for other tracks too (for example, this one could be operator track) 14:39:05 <rosmaita> i haven't had time to brainstorm a solution yet 14:39:19 <nikhil> cool 14:39:26 <rosmaita> yes, it would be good to have operator input for sure 14:39:53 <nikhil> I do have _some_ reservations around lifecycle management using metadata as properties can be ephemeral protected or not 14:40:24 <rosmaita> nikhil: i agree, that's why i paired these two topics together 14:40:27 <rosmaita> namely 14:40:38 <nikhil> gotcha 14:40:42 <rosmaita> expanding "common image properties" to be really common image properties 14:40:51 <rosmaita> and then maybe these lifecycle things would be one of those 14:41:05 <nikhil> hm, that's good thought 14:41:13 <rosmaita> it will help with import, too 14:41:25 <rosmaita> have specific image properties that are required 14:41:35 <nikhil> +1 14:41:37 <rosmaita> it's an interoperability thing, really 14:41:46 <nikhil> in general, it would be good to think how we can provide more context to this proposal so that rest of the clouds who may not need aren't really affected 14:42:12 <nikhil> but if you're thinking interop then may be all are involved 14:42:32 <rosmaita> so all clouds for topic (a), maybe not for topic (b) 14:42:41 <nikhil> I see 14:43:07 <rosmaita> metadefs gives us some tools we didn't have back in grizzly for common image properties 14:43:50 <nikhil> cool 14:44:12 <nikhil> rosmaita: I'm assuming you want to discuss (a) later? 14:44:31 <nikhil> or that you tied those up together in that problem statement? 14:44:45 <rosmaita> i could discuss now 14:45:03 <nikhil> sure, we've about 5 mins for this one :) 14:45:36 <rosmaita> the description is kind of self-explanatory 14:46:23 <rosmaita> so maybe no discussion? anyone have any questions? or it metadata too boring? 14:46:33 <rosmaita> or am i netsplit? 14:47:05 <nikhil> nah 14:47:20 <nikhil> I think it's one of those that may be good to discuss with other teams 14:47:30 <nikhil> particularly Nova in this case 14:47:33 <rosmaita> yes, nova should have some opinions 14:47:43 <nikhil> I want to know what other teams are using 14:48:00 <nikhil> and hopefully this feedback will help with the nova's compat layer conversation 14:48:22 <rosmaita> ok, well maybe discuss further on april 12 14:48:36 <nikhil> Thanks rosmaita! 14:48:38 <nikhil> moving on 14:48:53 <nikhil> #info Proposal #13 Deprecate glance-registry 14:49:15 <nikhil> flaper87: proposed this one, but I guess he's not here today 14:49:40 <nikhil> I think we need to give this one another 20 mins on april 12 14:49:51 <rosmaita> i became interested in it while working on a bug, and realized that the registry is a PITA 14:49:58 <nikhil> :) 14:49:59 <nikhil> One initial feedback from me: 14:50:26 <nikhil> We need to figure out which other services are using registry. for example, scrubber 14:50:36 <nikhil> and who all prefer that 14:50:59 <nikhil> may be I can replace services with tools above 14:51:15 <nikhil> anyone have comments? 14:51:30 <rosmaita> +1 to discuss on april 12 14:51:35 <nikhil> thanks 14:51:36 <mfedosin> yep 14:51:39 <nikhil> moving on 14:51:41 <nikhil> #info Proposal #12 Glance + Microversions 14:52:00 <nikhil> Well, I am interested to gather feedback 14:52:18 <nikhil> But I don't think we'd do this in parallel with other important API changes 14:52:24 <rosmaita> i agree it's worth looking into, but maybe not now 14:52:43 <nikhil> anyone else? 14:53:09 <nikhil> moving on 14:53:19 <nikhil> #info Proposal #11 Glance Store 1.0.0 Let's get the re-factor done 14:53:27 <nikhil> again, flaper87's proposal 14:53:45 <nikhil> we can just give him 45 mins slot on april 12 to discuss these items 14:53:58 <mfedosin> nikhil: frankly speaking refactor is needed 14:54:09 <mfedosin> because Glare is going to use glance_store 14:54:17 <nikhil> ok 14:54:33 <nikhil> mfedosin: but do you think refactor can be done in that small amount of time? 14:54:37 <mfedosin> and it doesn't satisfy our requirements 14:54:52 <nikhil> I see 14:54:59 <mfedosin> no, we have a layer that hides all things 14:55:17 <mfedosin> but it's better to refactor glance_store's code 14:55:23 <nikhil> well, there are some CP discussions around backward compatibility scope 14:55:45 <nikhil> I think it would be worth for you to be at that CP session on Tuesday 14:55:45 <mfedosin> and make it usable for all services not only Glance 14:56:04 <mfedosin> what time again? 14:56:11 <rosmaita> 1400-1800 utc 14:56:23 <rosmaita> oops, that's not it 14:56:46 <nikhil> yep, let's discuss the rest of the sessions next week on tuesday and those who don't get time can be discussed on thursday's meeting 14:56:55 <nikhil> rosmaita: that's correct 14:57:04 <rosmaita> oh, ok 14:57:15 <nikhil> Tuesday April 12, 1400-1800 UTC 14:57:22 <rosmaita> (i need 1 min of "general discussion" before meeting ends) 14:57:25 <mfedosin> not perfect for me but not bad 14:57:26 <nikhil> sure 14:57:33 <nikhil> #topic open discussion 14:57:37 <rosmaita> following up on my action item from last week, i put up a patch to the glance in-tree docs about "the future" of glance 14:57:43 <rosmaita> #link https://review.openstack.org/#/c/301359/ 14:57:53 <nikhil> mfedosin: there's a email thread if you want to provide feedback 14:57:54 <rosmaita> please take a look, especially cores, so we can all be on the same page before the summit 14:58:01 <nikhil> but I'd say a bit too late to change the time 14:58:06 <rosmaita> a few people have made some good suggestions about restructuring the glance in-tree docs, but at this point i'd really like to concentrate on content 14:58:16 <rosmaita> the aim of this exercise is to make sure that we're in agreement (more or less) about both (a) where the glance project should be going, and (b) some text that says where the glance project is going that we can point people to 14:58:27 <rosmaita> that's it, thanks! 14:58:45 <nikhil> Thanks rosmaita! This is really useful effort, especially prior to summit. 14:59:38 <nikhil> #info we may have to skip meeting on thursday april 21 so, please be ready next week with any important discussions 15:00:08 <mfedosin> we can skip april 28 too 15:00:08 <nikhil> That's it for today. Thanks all for joining. Cya on Tuesday. 15:00:17 <mfedosin> thanks! bye 15:00:27 <nikhil> connection info on the virtual sync will be on #openstack-glance ~1400 UTC 15:00:32 <nikhil> #endmeeting